True and false message identification method based on reinforcement learning and affair knowledge graph

Abstract

The invention provides a true and false message identification method based on reinforcement learning and a affair knowledge graph. The development of the internet and the new media breaks through the space-time limitation of network messages, so that false messages can generate huge influence on the real world in a short time. At present, rumor detection mainly depends on user reporting, information in a network cannot be detected on a large scale, and a detection method based on deep learning is low in timeliness and lacks training data. In order to improve the detection efficiency of true and false information in a network and timely perform true and false discrimination on emergencies, the invention provides a true and false message discrimination method based on reinforcement learning and a affair knowledge graph, a weak classifier is used for performing true and false classification on public opinion data, and the public opinion data with high confidence is screened by using reinforcement learning, so that the detection efficiency is improved. And carrying out authenticity identification on the screened public opinion data by using authenticity based on a affair knowledge graph, and updating reinforcement learning and a weak classifier according to an identification result.

Description

technical field [0001] The present invention relates to technologies such as deep learning, reinforcement learning, natural language processing, and knowledge graph of affairs, and specifically relates to a true and false message detection algorithm. Background technique [0002] At present, rumor detection mainly relies on user reports, and it is impossible to detect information in the network on a large scale. However, the detection method based on deep learning is not time-sensitive and lacks training data. In order to improve the detection efficiency of true and false information in the network and timely distinguish the true and false of emergencies, a method for identifying true and false information is proposed based on reinforcement learning and event knowledge graphs. The techniques closest to the present invention are: [0003] (1) Multimodal rumor detection method based on neural network: This method uses VGG-19 network to extract image content features, uses Den...

AI Technical Summary

Problems solved by technology

[0007] A true and false news identification model based on reinforcement learning and event knowledge graph, which stores the knowledge in public opinion through the event knowledge graph, solves the problem of difficult labeling of its data sets through reinforcement learning, and improves the generalization ability of the model

Images